Beruflich Dokumente
Kultur Dokumente
VOL. 38,
NO. 2,
MARCH/APRIL 2012
INTRODUCTION
. A. Sen is with the Department of Information and Operations Management, Mays Business School, Texas A&M University, College Station,
TX 77843. E-mail: asen@mays.tamu.edu.
. K. Ramamurthy and A.P. Sinha are with the Sheldon B. Lubar School of
Business, University of Wisconsin-Milwaukee, PO Box 742, Milwaukee,
WI 53201-0742. E-mail: {ramurthy, sinha}@uwm.edu.
Manuscript received 28 Apr. 2009; revised 24 May 2010; accepted 23 Oct.
2010; published online 3 Jan. 2011.
Recommended for acceptance by H. Muller.
For information on obtaining reprints of this article, please send e-mail to:
tse@computer.org, and reference IEEECS Log Number TSE-2009-04-0091.
Digital Object Identifier no. 10.1109/TSE.2011.2.
0098-5589/12/$31.00 2012 IEEE
attained the high levels of maturity that software development has, resulting in failed DW implementations, poor
data quality, and other associated problems.
Most companies embark on data quality initiatives to
address concerns such as spiraling direct-mail costs, poor
customer service, and faulty reports [54]. Poor data quality
costs money in terms of lost productivity, faulty business
decisions, and an inability to achieve results from expensive
investments in enterprise applications. One of the major
reasons for data quality problems is inconsistent data
definition. A data warehouse, which reflects the single
version of the truth [25] for an organization, is a prime touch
point for addressing data quality problems. Subject matter
experts who are knowledgeable about business as well as
data are often employed to define data cleansing rules and
data quality metrics, as well as recommend whether to fix the
data at the source, the staging area, or the warehouse [10].
In addition to addressing data quality concerns, a mature
DWP can be expected to provide several other benefits. A
mature DWP would help the organization to define and
deliver projects with predictable durations. It would force
the organization to develop data quality and data governance strategies that enhance the trust of sponsors and users
in the data. By building trust and providing the ability to
perform sophisticated business analytics, a mature DWP
would also keep the user base satisfied, thereby addressing
one of the main causes of data warehouse failures.
The need for a CMM-like maturity model for DWP has
been mooted in the literature [33], [34], [35], [60], [69]. But
none of the prior studies has gone beyond presenting a
sketchy or preliminary model. In this paper, we describe the
design and development of a comprehensive, detailed, and
robust DWP maturity modelbased on design-science
research guidelinesover a period of three years.
It is important to note that while there are many issues
common to software development and data warehousing,
there are a number of factors that render DWP unique. In
particular, any discussion on DWP maturity revolves
around data quality management, ETL design, metadata
management, data change management, data warehouse
governance, end-user cube design, etc.activities that do
not fall under the purview of traditional software development. The main contribution of our work is in designing a
DWP maturity model by identifying, defining, and accommodating those aspects that pertain specifically to DW.
The paper is organized as follows: Section 2 provides
the motivation for developing the DWP maturity (DWP-M)
model. Section 3 reviews the extant literature on maturity
models. Section 4 first presents the framework that has
been recently proposed for conducting design-science
research in IS, and then describes how we employed this
framework to design, develop, and evaluate the DWP-M
model. Section 5 discusses the contributions of our study
and Section 6 concludes the paper and identifies the future
research directions.
337
338
VOL. 38,
NO. 2,
MARCH/APRIL 2012
TABLE 1
Comparison of Maturity Models from Different Domains
abstraction attribute illustrates the intrinsic maturity paradigm that is used to develop the model. The model can
focus on either process maturity or product maturity.
Process maturity concentrates on how a process evolves
toward maturity. Product maturity, on the other hand,
explains the evolution to better products. The model benefits
attribute describes the benefits obtained from a maturity
model. The model scope attribute describes the focus of the
maturity model. The related technologies attribute portrays
the technologies that are covered by the maturity model.
Finally, the stakeholders attribute describes the people or
groups affected by the maturity exercise.
Our analysis of the extant models reveals that maturity
models focus on process or product evolution. All models,
directly or indirectly, use Crosbys framework to abstract
their inherent processes even though the stages of evolution
can differ. The process maturity modelssuch as IT Service
CMM [42], [43], [57], IQ CMM [12], and NASCIOs EMM
[41]borrow heavily from CMM. Some, such as the project
management maturity model [24], [39], have their own
specific models since CMM is geared toward IS companies.
Others, such as OSMM [16] and DW Maturity Model [11],
follow the Chasm Model espoused by Moore [38]. Also, note
that Eckersons DW maturity model [10], [11] focuses on
product (data warehouse) maturity, not on process maturity.
Even though many types of maturity models have been
proposed for different contexts (see Table 1), they all share a
set of common features. We find that a maturity model
typically supports the three key features described below:
Feature-1: Maturity levels. The idea of levels originated
from Crosbys work. The number of levels in a model
typically ranges from three to six. Each level usually has a
descriptor that serves as a name for the level. The Capability
Maturity Model developed by the Software Engineering
Institute (SEI) for software development process reflects the
best practices in software development and emphasizes the
need to conduct periodic software process assessments and
introduce improvements. CMM advocates that continuous
process improvement be based on small, evolutionary steps
and provides a framework for organizing those steps into
five maturity levels [50]. The five levels are initial, repeatable,
defined, managed, and optimizing.
Many organizations provide IT services [51], [70] either
internally or externally. These services include software
maintenance, operating information systems/data centers,
running networks, and providing technical support. Customers of these services at times may not be able to express
their real service requirements and may not know the
performance needs. Quite often, service providers also do
not know how to assess their own capabilities with respect
to the delivery of IT services. According to Niessink et al.
[43, p. 12], Regardless of the exact circumstances in which
an IT service provider operates, sufficient emphasis should
be on processes...to be able to deliver quality IT service. To
address these problems, Niessink and van Vliet [42] and
Niessink et al. [43] proposed the IT Service CMM model
based on the CMM Version 1.1 framework. The scope of
this model covers all of the service delivery activities and
focuses on the maturity of the service organization. Like
CMM, this model also has five levels and does not measure
339
340
VOL. 38,
NO. 2,
MARCH/APRIL 2012
341
342
2.
3.
VOL. 38,
NO. 2,
MARCH/APRIL 2012
343
TABLE 2
Data Warehousing Maturity Levels and KPAs (FIRST Version-June 2004)
344
VOL. 38,
NO. 2,
MARCH/APRIL 2012
KPA 4.6 covers ITILs 2e; KPA 3.7 covers ITILs ICT
infrastructure management; and KPA 3.17 covers ITILs
software asset management.
The DWP-M model, however, is substantially different
from ITIL. First, its focus is on both development and
service, unlike ITIL, which focuses primarily on service
activities. Second, our model places heavy emphasis on
continuous improvement, including prevention (in maturity level 5), while some research suggests that ITIL needs
to be complemented by six-sigma techniques to bring an
engineering orientation and lean techniques to promote
continuous improvement [72]. Third, the focus of our model
is exclusively on the data warehousing process, rather than
on IT governance in general. Finally, unlike our DWP-M
model, which requires a clear progression through the five
maturity levels and associated practices, ITIL can be
implemented on an as-needed basis, focusing on those
parts of the IT service delivery and management processes
that are broken [46].
In the new version of the model, the number of KPAs
increased from 29 to 40, and the total number of activities
covering those KPAs increased from 157 to 221, focusing on
both development and operations/customer service aspects. More than the increase in the number of KPAs and
activities, the scope of the model expanded to include DW
operations and services. There were a number of KPAs that
were common across these two aspects. At this point in
time, it was also necessary to identify and decide which
category of stakeholders would need to interact with the
DW teams in the real world in light of the different
emphasesdevelopment and operations/support services.
A fourth workshop was organized in March 2005.
Invitations were extended to the same companies that had
participated in the three earlier workshops. This was the
first time that the entire group of participants would see a
very different model with two different emphases and a
number of new KPAs.
Based on their comments and critiques, a fifth version
(version 1.5) was developed and presented to the panel of
DW experts via e-mail in the summer of 2005. The
participants were asked to propose, discuss, and arrive at
a consensual assignment of each KPA to DW development
staff and/or DW Operations Staff, and identify their
connections with the Business Users.
About 60 percent of the participants responded with
comments and critiques on version 1.5 of the model. After a
major cleanup, we created a sixth version (version 2.0) and
organized a fifth workshop in September 2005. Eight of the
original 13 companies and 12 experts from these companies
participated this time. The group session, as before, lasted
for about two and a half hours. The participants again
engaged in brainstorming to generate additional KPAs/
activities and ideas with respect to KPA assignment.
Based on the feedback, we revised the model. A seventh
version (version 2.1) was e-mailed to the participants. We
received individual feedback from a subset of the
participants this time. We created another revised DWPM Model (version 3.0) and sent out invitations for a sixth
workshop in March 2006. Nine companies and 10 experts
from these companies participated. Version 3.0 of the
DWP-M Model was presented to the group of experts
with a request to examine each maturity level and the
345
346
VOL. 38,
NO. 2,
MARCH/APRIL 2012
Fig. 1. Relationships among KPAs in DWP maturity model (partial): (a) Relationships in level 2. (b) Relationships in levels 3 and 4.
.
.
.
.
TABLE 3
FINAL Version of Data Warehousing Maturity Levels and KPAs (Distinguishing Development and Operations)
347
348
VOL. 38,
NO. 2,
MARCH/APRIL 2012
Fig. 2. Data warehousing process maturity model and KPAs (FINAL version).
All three experts were provided with detailed documentation of the DWP-M model. The experts were asked to
respond to each of the 25 statements on a 5-point Likert
scale, with 1 being strongly disagree and 5 being
strongly agree. To guard against mechanistic responses,
several questions were framed in the negative (e.g., The
DWP-M model is not helpful in assessing my organizations DW
process maturity). We determined the interrater reliability
among the three raters using the intraclass correlation
coefficient [37] using SPSS. The average measure for the
DISCUSSION
349
350
VOL. 38,
NO. 2,
MARCH/APRIL 2012
ACKNOWLEDGMENTS
The authors would like to thank the workshop panel
members for their active participation, support, and feedback during the development of the DWP maturity model.
The authors would also like to thank Dr. Hausi Muller and
the three anonymous referees for their insightful and
constructive reviews, which helped improve the quality of
the manuscript significantly.
REFERENCES
[1]
351
352
VOL. 38,
NO. 2,
MARCH/APRIL 2012
[69] H. Watson, T. Ariyachandra, and R.J. Matyska Jr., Data Warehousing Stages of Growth, Information Systems Management,
vol. 18, no. 3, pp. 42-51, Summer 2001.
[70] R.T. Watson, L.F. Pitt, and C.B. Kavan, Measuring Information
Systems Service Quality: Concerns for a Complete Canvas, MIS
Quarterly, vol. 21, no. 2, pp. 209-221, 1998.
[71] Wikipedia, Data Warehouse, http://en.wikipedia.org/wiki/
Data_warehouse, 2011.
[72] Wikipedia Information Technology Infrastructure Library,
http:/en.wikipedia.org/wiki/Information_Technology_Library,
May 2010.
[73] R. Winter and R. Burns, Managing Data Warehouse Growth,
Intelligent Enterprise, www.intelligententerprise.com/showArticle.
jhtml?articleID=193105574, Nov. 2006.
[74] B.H. Wixom and H.J. Watson, An Empirical Investigation of the
Factors Affecting Data Warehousing Success, MIS Quarterly,
vol. 25, no. 1, pp. 17-41, Mar. 2001.
[75] R.K. Yin, Case Study Research: Design and Methods, third ed. Sage
Publications, 2002.
Arun Sen received the MTech degree in
electronics from the University of Calcutta, India,
and the MS degree in computer science and the
PhD degree in information systems from Pennsylvania State University. He is a professor in
the Department of Information and Operations
Management, Mays Business School, Texas
A&M University. He has published more than
45 research papers in journals such as MIS
Quarterly, Information Systems Research, IEEE
Transactions on Software Engineering, IEEE Transactions on Systems,
Man, and Cybernetics, IEEE Transactions on Engineering Management,
Decision Sciences, Communications of the ACM, Information Systems,
Computers and OR, Omega, European Journal of Operational
Research, Decision Support Systems, Journal of Management Information Systems, Information and Management, and Omega. He has
served as an associate editor of the Journal of Database Management.
He has also been an editor of special issues for Decision Support
Systems, Communications of the ACM, Database, and Expert Systems
with Applications. He was the chair of the INFORMS College on
Information Systems, and a program chair for the Workshop on
Information Technologies and Systems (WITS) in 1996. His research
interests include data warehouse maturity, decision support systems,
database management, repository management and software reuse,
case- based reasoning, and e-Commerce.
353
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.